Proposed ad hoc Routing Approaches Conventional wired-type schemes (global routing, proactive): –Distance Vector; Link State Proactive ad hoc routing: –OLSR, TBRPF On- Demand, reactive routing: –DSR (Source routing), MSR, BSR –AODV (Backward learning) Scalable routing : –Hierarchical routing: HSR, Fisheye –OLSR + Fisheye – LANMAR (for teams/swarms) Geo-routing: GPSR, GeRaF, etc –Motion assisted routing –Direction Forwarding

On-Demand Routing Protocols Routes are established “on demand” as requested by the source Only the active routes are maintained by each node Channel/Memory overhead is minimized Two leading methods for route discovery: source routing and backward learning (similar to LAN interconnection routing)

Existing On-Demand Protocols Dynamic Source Routing (DSR) -- CMU Multipath Source Routing (MSR) – TJU Backup Source Routing (BSR) – UofO+TJU Ad-hoc On-demand Distance Vector (AODV) Associativity-Based Routing (ABR) Temporarily Ordered Routing Algorithm (TORA) Zone Routing Protocol (ZRP) Location assisted routing (LAR, DREAM) Signal Stability Based Adaptive Routing (SSA) On Demand Multicast Routing Protocol (ODMRP) – UCLA

Dynamic Source Routing (DSR) RFC 4728 – February 2007 Forwarding: source route driven instead of hop-by-hop route table driven –Mobility ? On-demand: No periodic routing update message is sent –Control traffic is minimized The first path discovered is selected as the route Two main phases (On-demand) –Route Discovery –Route Maintenance

DSR - Route Discovery Route RequestTo establish a route, the source floods a Route Request message with a unique request ID The Route Request packet “picks up” the node ID numbers Route ReplyRoute Reply message containing path information is sent back to the source either by –the destination, or –intermediate nodes that have a route to the destination Route CacheEach node maintains a Route Cache which records routes it has learned and overheard over time

DSR - Route Maintenance Route maintenance performed only while route is in use Monitors the validity of existing routes by passively listening to acknowledgments of data packets transmitted to neighboring nodes –acknowledgments of packets -- > contention ? (adaptive ack) Route ErrorWhen problem detected, send Route Error packet to original sender to perform new route discovery

MSR - Multipath Source Routing Direct Descendant of DSR On-demand + Source Routing + Multipath Probing-based adaptive load balancing among multiple paths Motivation of MSR –Efficiently using the network resource –Alleviate the oscillation in adaptive single path routing –Fast re-routing –Reducing computing & storage requirement –Exploiting computing power of host instead of link capacity

MSR Implementation To make multipath network viable –appropriate paths calculation –efficient packet forwarding on calculated paths –effective end-host usage of multiple paths In MSR –Path Finding –Packet forwarding –Load balancing

Distributing Traffic among Multiple Paths Quantities: A heuristic equation Probing-based adaptive control –Decoupling between transport layer and network layer: SRPing –Cost effective Scheduling: Packet Weighted Round Robin TCP out-of-order (re-sequencing) problem

Distributing Traffic among Multiple Paths Heuristic equation –Rationale: Autonomous system, homogeneous assumption, bandwidth-delay product constant where, is the delay of route with index i, is the maximum delay of all the routes to the same destination, R is a factor to control the switching frequency between routes. U is a bound value to insure that should not to be too large.

Simulation Methodology ns – Wireless extensions by CMU Adopt methods used in [Broch98, Johnson99] Two major files: –Movement pattern file –Communication pattern file 50 mobile hosts placed randomly within a 1500m×300m area Two set of simulations: Max speed 20m/s & 1m/s Different traffic types: CBR & FTP

Performance Evaluation MSR vs. DSR Performance Metrics –Packet delivery ratio –Data throughput –End-to-end delay –Packet drop probability –Queue size

Simulation Results – CBR Packet delivery ratio of every connection

Simulation Results – CBR Packet delivery ratio

Simulation Results – CBR End-to-end throughput

Simulation Results – TCP End-to-end throughput

Simulation Results – CBR End-to-end delay

Simulation Results – CBR End-to-end delay

Simulation Results – TCP End-to-end delay

Simulation Results - CBR Packets dropped at each node

Simulation Results – CBR IFQ queue size at each node

MSR Summary Reduce network congestion Improve throughput, delay, mobility, fault tolerance (CBR–more & FTP-less ? ) Acceptable routing overhead? –Little more than that of DSR –Route discovery –Route maintenance Probing (unicast) add little O/H Good candidate for QoS support –QoS-MSR, reliable-MSR Acceptable packet out-of-order level ? TCP

Backup Source Routing (BSR) Establish and maintain backup routes that can be utilized after the primary path breaks Define a new routing metric - backup route reliability E [T ,  ’] (analysis) –T ,  ’ -- the lifetime of backup route ( ,  ') –the basis for the backup path selection The lifetime for each link L in the network can be represented by an independent and identical (iid) exponential random variable X L Backup Route (  1,  2),  1 = (s, b, c, e, g, i, j, t) and  2=(s, a, b, c, d, f, h, i, k, t)

Backup Source Routing (BSR) Reduce the frequency of route discovery flooding, which is a major overhead in on- demand protocols Can improve the performance significantly in more challenging situations of high mobility Testbed (QoS) at TJU Multipath TCP with packet replicas (UCLA) –In a lossy ad hoc wireless network

Ad hoc On-Demand Distance Vector Routing (AODV) RFC3561 Primary Objectives –Provide unicast, broadcast, and multicast capability –Initiate forward route discovery only on demand –Disseminate changes in local connectivity to those neighboring nodes likely to need the information Characteristics –On-demand route creation Effect of topology changes is localized Control traffic is minimized –Distance vector routing –Two dimensional routing metric: Seq# -- to determine the freshness of the info –Storage of routes in Route Table

Route Table Fields: –Destination IP Address –Destination Sequence Number –to guarantee the loop-freedom of all routes towards that node –HopCount –Next Hop IP Address –Precursor Nodes –Expiration Time Each time a route entry is used to transmit data, the expiration time is updated to current_time + active_route_timeout Next Hop Source A Precursor Nodes Destination

Unicast Route Discovery <Flags, Bcast_ID, HopCnt, Src_Addr, Src_Seq#, Dst_Addr, Dst_Seq#> Node can reply to RREQ if –It is the destination, or –It has a “fresh enough” route to the destination Otherwise it rebroadcasts the request Nodes create reverse route entry Record Src IP Addr / Broadcast ID to prevent multiple rebroadcasts Source Destination Route Request Propagation Source broadcasts Route Request (RREQ)

Forward Path Setup Destination, or intermediate node with current route to destination, unicasts Route Reply (RREP) to source Nodes along path create forward route Source begins sending data when it receives first RREP Source Destination Forward Path Formation

Path Maintenance Movement of nodes not along active path does not trigger protocol action If source node moves, it can reinitiate route discovery When destination or intermediate node moves, upstream node of break broadcasts Route Error (RERR) message RERR contains list of all destinations no longer reachable due to link break RERR propagated until node with no precursors for destination is reached Source Destination 3’ Source Destination ’

Performance Evaluation Environment Simulation environment (QualNet) –100 nodes –1000mx1000m square area –transmission range: 100m –channel data rate: 2 Mbps –random mobility model –UDP traffic between randomly selected node pairs –cluster-token MAC layer protocol HSR –2 level physical partition –1 level logical groupings, number of logical subnets varies with network size FSR –2 level fisheye scoping –fisheye radius is 2 hops

Control O/H vs. number of nodes

Control O/H vs. Traffic Pairs

Control O/H vs. Mobility (100 pairs)

Average Delay (ms)

Location-Aided Routing (LAR) Ko and Vaidya (Texas A & M) Location assisted (requires GPS) On-demand No periodic messages Route RequestsLAR works like DSR except it limits the flooded area of Route Requests using location information

LAR (cont’d) Scheme 1 –The source specifies a request zone which includes the source and the area where the destination may reside Route Requests – Nodes within the request zone propagate Route Requests Scheme 2 –The source specifies the distance between itself and the destination Route Requests –Nodes forward Route Requests if their distances to the destination is less than or equal to the distance indicated by the packet

Distance Routing Effect Algorithm for Mobility (DREAM) Besagni, et al. (U of Texas, Dallas) Location assisted (requires GPS) Node coordinates (instead of routes) are recorded in the route table Distance EffectDistance Effect: Send location updates to nearby nodes more frequently Location update frequencies are adjusted to mobility rate

DREAM (cont’d) The source partially floods data to nodes that are in the direction of the destination The source specifies possible next hops in the data header using location information Next hop nodes select their own list of next hops and include the list into data header If the source finds no neighbors in the direction of the destination or has no fresh location information of the destination, data is flooded to the entire network

Location Based Routing Simulation (LAR and DREAM) 50 nodes; 750m X 750 m space Free space channel propagation model Radio with capture ability MAC: IEEE DCF 10 UDP data sessions with constant bit rate

Simulation Results (cont’d) Packet delivery ratio

Simulation Results Number of data packets transmitted per data packet delivered

Simulation Results (cont’d) Number of control bytes transmitted per data byte delivered

What have we discovered so far? Conventional (wired net) routing schemes (proactive routing) suffer of O/H, mobility & scalability limitations. On Demand routing eliminates background routing control O/H. It introduces latency; it does not well suited for QoS routing. Key observation: –Packet forwarding is done by routing tables (proactive and AODV) –Packet forwarding is done using the source route in header (DSR) –Thus, storage O/H; also route maintenance O/H Question: Can we eliminate table storage and maintenance O/H? Hierarchical routing reduces O/H and improves scalability (at the expense of accuracy).